30-08-2019, 02:15 PM
(This post was last modified: 30-08-2019, 09:49 PM by nickpelling.
Edit Reason: Added dshy instance count that had been omitted
)
Torsten: my proposed explanation for tightly bound pairs of letters begins with a specific, historically attested cipher mechanism - verbose cipher. This is a starting point that requires pairs of shapes to be locked together.
Your autocopying hypothesis, however, has no requirement for letters to be locked together. You have formulated a secondary explanation for this based on stroke harmony adjacency rules: which strokes can follow what other strokes determines, in your secondary system, what can be substituted for what.
Except... these secondary rules only work in those places where you want them to work - i.e. in transitions/replacements you think are 'justified' (because they appear in Voynichese), but not in places where you don't want them to work (because these don't / rarely appear in Voynichese).
An example: if o can happily be followed by r 2723 times, what is stopping other shapes going in between them?
or 2723
oar 85
olr 14
odr 5
oer 1
okr 1
Similarly for o and l:
ol 5488
odl 20
ool 19
okl 15
otl 9
orl 7
Similarly for d and y:
dy 6696
dchy 61
dey 15
dshy 12
dly 7
dky 5
dpy 4
doy 3
day 2
Only some of these glyph insertions are 'prohibited' by the stroke harmony adjacency rules: at the same time, to my eyes, many seem to be completely possible, yet have very low instance counts in the actual manuscript.
My conclusion is therefore that your (well, Elias Schwertfeger's) stroke harmony rules give a very incomplete account of the letter-to-letter statistics we actually see, and hence don't offer a workable explanation in and of themselves why certain glyph pairs are so tightly bound that they resist insertions so vigorously. For is not the idea of autocopying based around shape insertions and removals?
Your autocopying hypothesis, however, has no requirement for letters to be locked together. You have formulated a secondary explanation for this based on stroke harmony adjacency rules: which strokes can follow what other strokes determines, in your secondary system, what can be substituted for what.
Except... these secondary rules only work in those places where you want them to work - i.e. in transitions/replacements you think are 'justified' (because they appear in Voynichese), but not in places where you don't want them to work (because these don't / rarely appear in Voynichese).
An example: if o can happily be followed by r 2723 times, what is stopping other shapes going in between them?
or 2723
oar 85
olr 14
odr 5
oer 1
okr 1
Similarly for o and l:
ol 5488
odl 20
ool 19
okl 15
otl 9
orl 7
Similarly for d and y:
dy 6696
dchy 61
dey 15
dshy 12
dly 7
dky 5
dpy 4
doy 3
day 2
Only some of these glyph insertions are 'prohibited' by the stroke harmony adjacency rules: at the same time, to my eyes, many seem to be completely possible, yet have very low instance counts in the actual manuscript.
My conclusion is therefore that your (well, Elias Schwertfeger's) stroke harmony rules give a very incomplete account of the letter-to-letter statistics we actually see, and hence don't offer a workable explanation in and of themselves why certain glyph pairs are so tightly bound that they resist insertions so vigorously. For is not the idea of autocopying based around shape insertions and removals?